Lab3 数字图像分割与边缘检测实验¶
Author: Gao Pengbing (nbgao)
Email: nbgao@126.com
Work1 使用一阶和二阶检测算子（导数）进行图像的边缘检测¶
Work1-P1 编写程序实现一阶Sobel算子，进行图像的边缘提取
Work1-P2 编写程序实现一阶Prewitt算子，进行图像的边缘提取
Work1-P3 编写程序实现一阶Roberts算子，进行图像的边缘提取
Work1-P4 编写程序实现二阶Laplacian算子(3*3)进行图像的边缘提取
In [2]:
import numpy as np
import matplotlib.pyplot as plt
import skimage
import cv2
import warnings
%matplotlib inline
warnings.filterwarnings('ignore')
Sobel Operator¶
In [3]:
''' Sobel
Horizonal (X)
Gy =
-1 -2 -1
0 0 0
1 2 1
Vertical (Y)
Gx =
-1 0 1
-2 0 2
-1 0 1
'''
def SobelFilter(I, orientation):
[h, w] = I.shape[:2]
G = np.zeros((h, w))
# Filter
K = np.ones((3,3))
if(orientation in ['h', 'H', 'y', 'Y']): # Horizon Edge Detect
K = np.array([[-1,-2,-1],[0,0,0],[1,2,1]])
elif(orientation in ['v', 'V', 'x', 'X']): # Verticle Edge Detect
K = np.array([[-1,0,1],[-2,0,2],[-1,0,1]])
for i in range(1,h-1):
for j in range(1,w-1):
T = I[i-1:i+2, j-1:j+2]
G[i,j] = np.sum(T*K)
# G[i,j] = np.abs(np.sum(T*K))
# G[i,j] = 255 if G[i,j]>255 else G[i,j]
# G[G>255] = 255
# G = np.uint8(G)
# 取绝对值，大于255的置为255
G = cv2.convertScaleAbs(G)
return G
Prewitt Operator¶
In [4]:
''' Prewitt
Horizonal (X)
Gy =
-1 -1 -1
0 0 0
1 1 1
Vertical (Y)
Gx =
-1 0 1
-1 0 1
-1 0 1
'''
def PrewittFilter(I, orientation):
[h, w] = I.shape[:2]
G = np.zeros((h, w))
# Filter
K = np.ones((3,3))
if(orientation in ['h', 'H', 'y', 'Y']): # Horizon Edge Detect
K = np.array([[-1,-1,-1],[0,0,0],[1,1,1]])
elif(orientation in ['v', 'V', 'x', 'X']): # Verticle Edge Detect
K = np.array([[-1,0,1],[-1,0,1],[-1,0,1]])
for i in range(1,h-1):
for j in range(1,w-1):
T = I[i-1:i+2, j-1:j+2]
G[i,j] = np.sum(T * K)
G = cv2.convertScaleAbs(G)
return G
Roberts Operator¶
In [5]:
'''
Gx =
1 0
0 -1
Gy =
0 1
-1 0
'''
def RobertsFilter(I, orientation):
[h, w] = I.shape[:2]
G = np.zeros((h, w))
# Filter
K = np.ones((2,2))
if(orientation in ['l', 0]):
K = np.array([[1, 0], [0, -1]])
elif(orientation in ['r', 1]):
K = np.array([[0, 1], [-1, 0]])
for i in range(0, h-2):
for j in range(0, w-2):
T = I[i:i+2, j:j+2]
G[i,j] = np.sum(T*K)
G = cv2.convertScaleAbs(G)
return G
Laplacian Operator¶
In [6]:
''' Laplacian
small
K =
0 -1 0
-1 4 -1
0 -1 0
large
K =
-1 -1 -1
-1 8 -1
-1 -1 -1
'''
def LaplacianFilter(I, mode):
[h, w] = I.shape[:2]
G = np.zeros((h, w))
# Filter
K = np.ones((3,3))
if(mode in ['4', 's', 4]):
K = np.array([[0,-1,0],[-1,4,-1],[0,-1,0]])
elif(mode in ['8', 'l', 8]):
K = np.array([[-1,-1,-1],[-1,8,-1],[-1,-1,-1]])
for i in range(1,h-1):
for j in range(1,w-1):
T = I[i-1:i+2, j-1:j+2]
G[i,j] = np.sum(T * K)
G = cv2.convertScaleAbs(G)
return G
In [7]:
for i in range(1,6):
file_path = '../Image/image'+str(i)+'.jpg'
img = plt.imread(file_path)
img_gray = np.uint8(skimage.color.rgb2gray(img)*255)
img_sobel_x = SobelFilter(img_gray, 'x')
img_sobel_y = SobelFilter(img_gray, 'y')
img_prewitt_x = PrewittFilter(img_gray, 'x')
img_prewitt_y = PrewittFilter(img_gray, 'y')
img_roberts_l = RobertsFilter(img_gray, 'l')
img_roberts_r = RobertsFilter(img_gray, 'r')
img_laplacian_4 = LaplacianFilter(img_gray, '4')
img_laplacian_8 = LaplacianFilter(img_gray, '8')
plt.figure(figsize=(12, 4*6*img.shape[0]/img.shape[1]))
plt.subplot(421)
plt.imshow(img_sobel_x, 'gray')
plt.title('sobel_x')
plt.axis('off')
plt.subplot(422)
plt.imshow(img_sobel_y, 'gray')
plt.title('sobel_y')
plt.axis('off')
plt.subplot(423)
plt.imshow(img_prewitt_x, 'gray')
plt.title('prewitt_x')
plt.axis('off')
plt.subplot(424)
plt.imshow(img_prewitt_y, 'gray')
plt.title('prewitt_y')
plt.axis('off')
plt.subplot(425)
plt.imshow(img_roberts_l, 'gray')
plt.title('roberts_l')
plt.axis('off')
plt.subplot(426)
plt.imshow(img_roberts_r, 'gray')
plt.title('roberts_r')
plt.axis('off')
plt.subplot(427)
plt.imshow(img_laplacian_4, 'gray')
plt.title('laplacian_4')
plt.axis('off')
plt.subplot(428)
plt.imshow(img_laplacian_8, 'gray')
plt.title('laplacian_8')
plt.axis('off')
plt.show()
